DESCRIPTION: Screening mammography for early detection of breast cancer improves the chances of cure and allows for less traumatic and less expensive treatment. Studies have shown that screening suffers from large variability in detection rates, and that radiologists do not identify all breast cancers that are visible upon retrospective review. The ultimate goal of the proposed project is to develop and market intelligent software for mammogram image analysis to assist radiologists in detecting early stage breast cancer. The software will prompt the radiologist to more carefully study computer-detected suspicious regions in the image which has been shown to improve detection rates of breast cancer. The first specific aim of phase I is to combine previously developed software components which address subparts of the automated detection problem into a complete prototype capable of detecting abnormalities found in a screening environment. This will be performed by company researchers who have extensive experience in this field. The second aim is to finalize clinical trial design for clinical testing to be carried out in phase II. This will be accomplished in collaboration with expert radiologists and an experienced biostatistician. Clinical trials are necessary for Food and Drug Administration approval and future commercialization of the prototype.